Research On Intelligent Lane-changing And Exit Strategy Considering Dynamic Dangerous Conditions

Lane-changing was a routine driving operation.Due to blind areas and misoperations during the lane-changing,it was easy to cause collision accidents.Therefore,it was necessary to develop the lane-changing Advanced Driving Assistance System to improve safety.Starting from the algorithm of the lane-changing Advanced Driving Assistance System,the manuscript established a research on intelligent lane-changing and exit strategies considering dynamic dangerous conditions,ensure vehicles can exit from lane-changing when encountering danger.The classification model was established including danger level based on other vehicles’ intention recognition,the whole process lane-changing safety model,the trajectory planning model with exiting mechanism,the verification model of lane-changing trajectory tracking control based on predictive control.Firstly,the classification model was established based on other vehicles intention recognition.Four working conditions were constructed according to the location of other vehicles.The results of the intention recognition were classified according to the four working conditions.The safe type would continue to change lanes,and the dangerous type would exit lane-changing.Uncertain types were input into the whole process lanechanging safety model for further judgment.The Gauss-Hidden Markov model was used for intention recognition,three-channel data was collected through a driving simulator,the data was divided into two section for training and verifying.Using the data which from training to train the model and established an online intent recognition model.The virtual scene verification showed that the intent of another car was recognized at 1.1s.The verification data showed that the recognition accuracy was above 96%,and the average recognition time was less than 1.2s,verifying the accuracy and rapidity of the model.Secondly,the whole process safety model considering the dynamic dangerous conditions of other vehicles was established.The safety model was the judgment for lanechanging exiting.Calculated the critical lane-changing collision time under the four working conditions considering the dynamic change,and predicted the arrival time of the other vehicle according to the current speed and acceleration,and then obtained the lanechanging collision time.Establish a four-condition parallel full-process lane-changing safety model.A fuzzy logic solution to the dynamic safety threshold time model was established.The dynamic safety threshold time was determined by considering vehicle speed,road conditions,and weather conditions adapting different environmental requirements.The scene verification was carried out by changing the speed of the vehicle and the sudden acceleration time,the accuracy of the safety model and fuzzy logic threshold time model were verified.The sudden acceleration value was obtained through the NGSIM data to simulate the actual vehicle to verify the safety model when the other vehicle changes dynamically.Later,a whole-process lane-changing trajectory planning model with exit mechanism was proposed for normal lane-changing and exit-lane-changing trajectory planning.When the lane-changing driving assistance system was turned on,the normal lane-changing trajectory planning was performed.The vehicle would exit the process of lane-changing,when the dangerous condition was enabled or the dangerous type was defined from the danger level classification model.Secondly,the objective function of the lane-changing trajectory constraint optimization was established.The constraint conditions mainly included the vehicle’s dynamic constraint and the safety distance constraint.An optimization objective function was established for normal lane-changing considering comfort and rapidity,at the same time the penalty item was added for optimization to ensure safety in lane-changing exit.Finally,the trajectory planning model was simulated and analyzed,the accuracy of the trajectory planning model and the multi-factor optimization model was verified by changing the weight coefficients of rapidity and adding penalty item.Finally,the trackability of the lane-changing trajectory was verified,the difference between the lane-changing trajectory and the actual tracking trajectory were detected.At the same time,a model prediction controller was built to track and control the trajectory during the lane-changing process.The vehicle speed was changed to verify the accurate performance in low-speed and high-speed conditions.At the same time,the speed adaptability of the lane-changing trajectory tracking controller was analyzed.The results showed that the control system had strong speed adaptability.The rationality and traceability of the lane-changing trajectory under different speed conditions were verified.